
Who: Giacomo Luani, PhD candidate at the Porous Media Physics section, Department of Physics, NTNU, Trondheim, Norway
When: Wednesday 1 October, between 13:00 and 14:00 CET (Norway time)
Where: The lecture will be in-person in the common room (PoreLab Trondheim) and streamed in the Kelvin room ( Porelab Oslo). From anywhere else, you will be able to join via the following Zoom link: https://uio.zoom.us/j/69096081290?pwd=bg1XKTMKjweeHqxM77VmAfbOLPD0o6.1
Title: Enabling Fast 4D X-ray CT in the home-lab with Machine Learning
Summary:
X-ray Computed Tomography (CT) is a well-established, non-destructive 3D imaging technique widely used across fields ranging from medical diagnostics to industrial inspection. Its micrometer-scale spatial resolution makes it an indispensable tool for studying porous media at the pore scale. Time-resolved CT (4D-CT) extends traditional 3D CT to capture dynamic processes such as multiphase flow. However, with conventional sampling and reconstruction methods, achieving both high image quality and high temporal resolution remains challenging—particularly in standard laboratory settings. Here, we introduce NeCT, a physics-informed deep learning model that delivers state-of-the-art performance in sparse-view and 4D-CT reconstruction. NeCT achieves temporal resolutions on the order of a few seconds while preserving the spatial resolution expected from conventional micro-CT systems. We demonstrate its capabilities through fast imaging of liquid imbibition in sandstone, where the high spatiotemporal resolution enables direct observation of salt dissolution and pore-filling events.
What is the PoreLab Early Career Lecture series ? The Porelab Early Career Lecture Series is a sub-category of the normal lecture series that gives Junior researchers (BSc, MSc, PhD students) and their promising research a podium. If you have a candidate for these series, please contact one of the organizers and submit your recommendations.